Presently, prefetching blindly downloads content which may or may not be utilized by a client system in the future. Such an operation is performed without any consideration for the probability that the object may actually be used, nor does prefetching take into consideration the size of the object, the bandwidth of the link between the client and the content server, etc. As a result, a considerable amount of bandwidth, server capacity, and storage space is wasted on prefetched content which is never actually used. Hence, improvements in the art are needed.